Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/cloudwaddie/lmarenabridge/llms.txt

Use this file to discover all available pages before exploring further.

LMArena Bridge is a self-hosted server that exposes an OpenAI-compatible API on top of LMArena. This guide walks you through cloning the project, getting your LMArena auth token, and making your first request.

​
Prerequisites

  • Python 3.10 or higher
  • pip (comes with Python)
  • A browser logged in to lmarena.ai or arena.ai

​
Install

1

Clone the repository

git clone https://github.com/CloudWaddie/LMArenaBridge.git
cd LMArenaBridge
2

Install dependencies

pip install -r requirements.txt
3

Start the server

python -m src.main
The server starts on http://localhost:8000. You should see startup output as the bridge initialises its browser session and scrapes model data from LMArena.

​
Get your auth token

LMArena Bridge authenticates to LMArena using the arena-auth-prod-v1 cookie from your browser session. Here is how to find it.
1

Open LMArena and send a message

Go to arena.ai (or lmarena.ai) and send any message in the chat. This ensures the auth cookie is set in your browser.
2

Open browser developer tools

Press F12 to open DevTools, or right-click the page and choose Inspect.
3

Navigate to the Cookies section

  • In Chrome or Edge: go to the Application tab, then expand Cookies in the left sidebar and select the LMArena domain.
  • In Firefox: go to the Storage tab, then expand Cookies.
4

Copy the arena-auth-prod-v1 value

Find the cookie named arena-auth-prod-v1. Its value starts with base64-. Copy the entire value — this is your auth token.
If you see cookies named arena-auth-prod-v1.0 and arena-auth-prod-v1.1 instead, copy both values and concatenate them end-to-end. The bridge handles split cookies automatically when you add tokens via the dashboard.

​
Add the token in the dashboard

1

Open the dashboard

Go to http://localhost:8000/dashboard in your browser.
2

Log in

Enter the default password: admin
Change this password immediately after first login. You can update it in the dashboard settings or by editing the password field in config.json.
3

Add your auth token

In the Auth Tokens section, paste the arena-auth-prod-v1 value you copied and click Add Token. The dashboard shows the token’s expiry date and whether it is currently active.

​
Make your first API call

The bridge is now ready to accept requests. Use the same format as the OpenAI API: 
curl http://localhost:8000/api/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": false
  }'
Replace YOUR_API_KEY with an API key from the dashboard. If you have not created one yet, you can leave the Authorization header out — or create a key in the API Keys section of the dashboard first. A successful response looks like this: 
{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "created": 1700000000,
  "model": "gpt-4o",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help you today?"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 10,
    "completion_tokens": 9,
    "total_tokens": 19
  }
}
To see all available models, call GET /api/v1/models. The bridge fetches the model list from LMArena on startup and refreshes it every 30 minutes.

​
Next steps

  • Authentication — Manage multiple auth tokens, API keys, and the cf_clearance cookie.
  • Configuration — Tune rate limits, timeouts, and browser window modes via config.json.
  • Integrations — Connect LMArena Bridge to OpenWebUI as a backend.

Build docs developers (and LLMs) love

Get started for freeTalk to us